From PI2 review chat re: ALMA PB implementation for technology demonstrator (#49)
10:22:16 From George Moellenbrock-NRAO : A question in the weeds: why is the PB difference image non-zero, and why does it appear to have a noisy component? And what are the radial features? I would have thought the PB images would be more analytical (and thus ~smooth), being a function of antenna properties and not of the data itself. Maybe I’ve misunderstood what is being shown.
10:28:44 From Eric Koch : Replying to "A question in the we..."
Tak’s example notebook noted that there’s a discrepancy in how the obscured Airy disk model is implemented: https://github.com/casangi/astroviper/blob/main/docs/core_tutorials/imaging/cube_single_field_primary_beam.ipynb
Tak, is that your understanding as the source for the small discrepancy?
10:29:14 From Srikrishna Sekhar : Replying to "A question in the we..."
Tak did a deep dive into this, I can let him answer more fully. But it seems to be because CASA adopts an approximation to the airy disk, but the exact cause of those radial patterns is not known yet (as far as I know…)
10:32:15 From ttsutsum : Replying to "A question in the we..."
The plot show the same airy model between CASA and this implementation but as Krishina pointed out, there are difference in sampling of the model and CASA uses interpolation to match to the image.
10:34:27 From Todd Hunter : I should point out here that the “angularly scaled” Airy disk itself is not the best representation of the ALMA primary beam, as the dishes are not uniformly illuminated, in fact much less so than VLA (and ngVLA). But it has been “good enough” for now at the pb>0.1 level (within 1% of a Gaussian illumination, in theory). Note that it will be very wrong in the sidelobes (many dB).
10:37:50 From Ryan Loomis : Replying to "A question in the we..."
@todd Hunter, we have captured this feedback in a github issue to be addressed in the future
10:57:59 From George Moellenbrock-NRAO : Replying to "A question in the we..."
I’m ok with the systematic differences (essentially net scaling). That there is pixel noise is mysterious.....
From PI2 review chat re: ALMA PB implementation for technology demonstrator (#49)
10:22:16 From George Moellenbrock-NRAO : A question in the weeds: why is the PB difference image non-zero, and why does it appear to have a noisy component? And what are the radial features? I would have thought the PB images would be more analytical (and thus ~smooth), being a function of antenna properties and not of the data itself. Maybe I’ve misunderstood what is being shown.
10:28:44 From Eric Koch : Replying to "A question in the we..."
Tak’s example notebook noted that there’s a discrepancy in how the obscured Airy disk model is implemented: https://github.com/casangi/astroviper/blob/main/docs/core_tutorials/imaging/cube_single_field_primary_beam.ipynb
Tak, is that your understanding as the source for the small discrepancy?
10:29:14 From Srikrishna Sekhar : Replying to "A question in the we..."
Tak did a deep dive into this, I can let him answer more fully. But it seems to be because CASA adopts an approximation to the airy disk, but the exact cause of those radial patterns is not known yet (as far as I know…)
10:32:15 From ttsutsum : Replying to "A question in the we..."
The plot show the same airy model between CASA and this implementation but as Krishina pointed out, there are difference in sampling of the model and CASA uses interpolation to match to the image.
10:34:27 From Todd Hunter : I should point out here that the “angularly scaled” Airy disk itself is not the best representation of the ALMA primary beam, as the dishes are not uniformly illuminated, in fact much less so than VLA (and ngVLA). But it has been “good enough” for now at the pb>0.1 level (within 1% of a Gaussian illumination, in theory). Note that it will be very wrong in the sidelobes (many dB).
10:37:50 From Ryan Loomis : Replying to "A question in the we..."
@todd Hunter, we have captured this feedback in a github issue to be addressed in the future
10:57:59 From George Moellenbrock-NRAO : Replying to "A question in the we..."
I’m ok with the systematic differences (essentially net scaling). That there is pixel noise is mysterious.....